Traditional pharmacology posits that a single compound acting through a single receptor will cause a single type of functional response (either full, partial or inverse agonism, or antagonism) for all effector pathways associated with that receptor and its milieu. Accordingly, compounds have been categorized by their ‘intrinsic efficacy’, often defined by the ability of a ligand to modulate receptor-mediated adenylate cyclase (AC) activity. There is increasing evidence, however, that many ligands do not conform to such a rigid definition of function. In fact, recent observations have led to a growing acceptance of the idea that one ligand, while acting on a specific receptor subtype, can have multiple intrinsic activities depending upon the effectors being examined and the model being employed (Mailman and Gay, 2004; Simmons, 2005).

For example, we have shown previously that a number of dopamine D2 ligands exhibit functionally selective profiles (Kilts et al, 2002; Mottola et al, 2002; Gay et al, 2004). Further evidence for the ability of ligands to activate G-protein-coupled receptor (GPCR)-mediated effectors differentially has been illustrated in serotonin 5-HT2A and 5-HT2C (Berg et al, 1998; Kurrasch-Orbaugh et al, 2003), α2A-adrenergic (Brink et al, 2000; Kukkonen et al, 2001), β2-adrenergic (Ghanouni et al, 2001), cannabinoid CB1 (Glass and Northup, 1999), μ-opioid (Allouche et al, 1999), and the oxytocin (Reversi et al, 2005) receptor expression systems, among others, effectively illustrating functional selectivity as a universal mechanism of GPCR effector regulation. It is also important to point out that functional selectivity is not an epiphenomenon of a specific receptor expression system, as all of the above observations were made in a number of different physiological models. This mechanism may be relevant to topical issues in neuropsychopharmacology.

Dysfunctional dopaminergic neurotransmission is considered a primary mechanism of schizophrenic symptomology (Kapur and Remington, 2001), indeed, the early pharmacotherapy for schizophrenia was based on serendipitous discovery of drugs that turned out to be dopamine receptor antagonists (Carlsson, 1964). Until recently, all of the antipsychotic drugs (APDs), whether ‘typical’ or ‘atypical’, or whether of high or low affinity, have been functional D2 antagonists (Miyamoto et al, 2000; Davies et al, 2004; Maudsley et al, 2005). A clear exception to this, however, is the recently approved compound aripiprazole (Abilify). The unique pharmacology of this compound first was demonstrated in models that showed aripiprazole-activated effectors associated with presynaptic D2 autoreceptors, whereas it antagonized D2 postsynaptic receptor-mediated effects (Kikuchi et al, 1995). Aripiprazole partially activated D2 receptor-mediated inhibition of cyclic adenosine 3′,5′-cyclic monophosphate (cAMP) accumulation, although this action was system specific (Lawler et al, 1999; Shapiro et al, 2003). It was suggested that aripiprazole appeared to activate D2 receptor-mediated effectors differentially, and could be termed a D2 functionally selective drug (Lawler et al, 1999; Shapiro et al, 2003). Conversely, Burris et al (2002) have proposed that the unique properties of aripiprazole result solely from its partial agonist properties.

The potential of partial D2 agonists as a novel treatment of schizophrenia was based in large part on data showing that apomorphine, a high-affinity D2 dopamine receptor agonist, could preferentially activate D2 autoreceptors at low doses (Tamminga et al, 1978; Roth, 1979). It was hypothesized that a D2 partial agonist might attenuate the activity of hyperactive mesolimbic neurons, and possibly increase neurotransmission in neurons where there was a deficit of activity (ie mesocortical neurons related to working memory). Presumably such compounds also would induce minimal extrapyramidal side effects (and tardive dyskinesia) that are associated with the receptor blockade caused by typical APDs. The potent D2-like receptor-selective agonist N-propylnorapomorphine (Tamminga et al, 1986) and (−)3PPP (S(−)-3-(3-hydroxyphenyl)-N-propylpiperidine hydrochloride, a D2 receptor partial agonist) thus seemed to have antipsychotic potential, although patients quickly lost beneficial effects of these agonists rather than having increased efficacy with time as with APDs (Clark et al, 1982; Tamminga et al, 1992; Lahti et al, 1998). For these reasons, there was a great deal of interest in aripiprazole based on the suggestion that it was the long-anticipated D2 partial agonist (Lawler et al, 1999; Burris et al, 2002; Shapiro et al, 2003). In the process, the idea of functional selectivity (Lawler et al, 1999), despite additional support for this mechanism (Shapiro et al, 2003), has been dismissed by many authoritative sources (Stahl, 2001; Tamminga, 2002; Lieberman, 2004).

Previous work has examined functional effects of aripiprazole at D2-regulated AC and receptor-regulated potassium channels (Lawler et al, 1999; Burris et al, 2002; Shapiro et al, 2003). The current study extended this by examining the binding characteristics of aripiprazole at the low and normal affinity states of the dopamine D2 receptor, as well as the effects of the drug on D2L receptor-mediated phosphorylation of mitogen-activated protein kinase (MAPK), potentiation of arachidonic acid (AA) release (Missale et al, 1998), and ligand-induced D2 receptor internalization.



The following were generously donated to this study: aripiprazole (OPC-14597) from Otsuka America Pharmaceuticals (Rockville, MD); olanzapine from Eli Lilly Inc. (Indianapolis, IN); melperone (from Cilag AG-Switzerland); and amisulpride from Dr Shitij Kapur (University of Toronto, Canada). Quinpirole, (−)3PPP, haloperidol, and clozapine were purchased from Sigma/RBI (Natick, MA), whereas dopamine, mepacrine, staurosporine, melittin, adenosine triphosphate, guanosine triphosphate, EDTA, dithiothreitol, sucrose, pepstatin A, leupeptin, PMSF, and other standard laboratory compounds were purchased from Sigma Chemical Co. (St Louis, MO). Sources of other reagents were as follows: 2′-amino-3′-methoxyflavone (PD98059) was purchased from BIOMOL Research Laboratories Inc. (Plymouth Meeting, PA); [3H]N-methylspiperone from Perkin-Elmer Life Sciences Inc. (Boston, MA); [5,6,8,9,11,12,13,14-3H]AA from Amersham Biosciences Inc. (Piscataway, NJ); 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) from Research Organics Inc. (Cleveland, OH); Ham's F-12 and DMEM media, penicillin, streptomycin, and geneticin (G418) from Invitrogen Co. (Carlsbad, CA); and primary antibody to phospho-Erk1/2 MAPK and secondary anti-rabbit HRP-conjugated antibody from Cell Signaling Technology Inc. (Beverly, MA). M1 anti-FLAG antibody was purchased from Sigma and Alexa594-conjugated goat anti-mouse was from Jackson ImmunoResearch (Malvern, PA).

Cells and Membranes

The Chinese hamster ovary cells (CHO)-hD2L cells are a stable line originally obtained from Dr Tony Sandrasagra (Aventis, Bridgewater, NJ). CHO-hD2L cells were maintained in Ham's F-12 medium supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 500 μg/ml G418 at 37°C and 5.0% CO2. CHO-hD2L cells were grown to confluence in 75-cm2 flasks. A measure of 5 ml of cold phosphate-buffered saline was used per flask to rinse the cells, after which 5 ml of lysis buffer (2 mM HEPES, 2 mM EDTA, 1 mM dithiothreitol, 1 μg/ml pepstatin A, 0.5 μg/ml leupeptin, and 10 μg/ml PMSF, pH 7.4 with HCl) was added to each flask. Human embryonic kidney cells (HEK)-FLAG-D2L cells were also grown and handled in a similar manner as the CHO-hD2L cells. Following 10–20 min of incubation at 4°C, the cells were scraped and collected. The cell suspension was homogenized with three strokes in a Wheaton glass homogenizer and centrifuged at 30 000g for 20 min. The resulting supernatant was discarded, the pellet resuspended in storage buffer (50 mM HEPES, 0.32 M sucrose, 1 μg/ml pepstatin A, 0.5 μg/ml leupeptin, and 10 μg/ml PMSF, pH 7.4 with NaOH) at approximately 1 mg protein/ml, homogenized again, and aliquoted into 1-ml microcentrifuge tubes. The cell membranes then were frozen, and stored at −80°C until further use.

Saturation and Competition Binding

Saturation studies were performed to verify the expression level of hD2L receptor in the CHO cells, as well as to determine the levels of FLAG-tagged D2L receptor expressing in the HEK cells. Membranes were incubated with varying concentrations of [3H]N-methylspiperone in binding buffer (50 mM HEPES, 4 mM MgCl2, pH=7.4, with KOH). Nonspecific binding was determined using 10 μM domperidone. Competition binding studies were carried out using 0.3 nM [3H]N-methylspiperone with and without 600 μM guanosine triphosphate nucleotide (GTP) to determine differences in affinity of each D2 receptor ligand for the high- and low-affinity hD2L receptor states in the CHO cell line. Total binding was determined by the amount of radioligand binding in the absence of competing drug, while nonspecific binding was defined by the amount of radioligand bound in the presence of 10 μM haloperidol. Seven to eight concentrations of each D2 receptor ligand were used. For both binding paradigms, addition of the tissue to each assay tube initiated the binding. Each drug condition was run in triplicate per experiment in a final volume of 500 μl. Following a 15 min incubation at 37°C, tubes were filtered through a FilterMate 196 Cell Harvester (Perkin-Elmer Life and Analytical Sciences, Boston, MA) and the plates were washed four times with ice-cold buffer. The filters were dried in an oven at 55°C for 30 min, and 35 μl of Packard MicroScint 20 scintillation cocktail was added to each well (Perkin-Elmer). A Packard TopCount NXT (Perkin-Elmer) was used to determine the radioactivity of each sample. Saturation binding data were expressed in fmol of receptor/mg of protein. Competition binding data were expressed as a percentage of specific binding.

Cell-Based ELISA for the Measurement of Mitogen-Activated Protein Kinase (MAPK) Activation

Measurements of phosphorylated MAPK were made using a published protocol (Versteeg et al, 2000). CHO cells (both wild type and those transfected with the human D2L receptor) were seeded in 96-well plates in Ham's F-12 media (10% FBS) at 50 000 cells/cm2 and allowed to grow at 37°C and 5% CO2 for 48 h. Cells were serum starved for 6 h prior to stimulation, after which appropriate drugs were added to each well at a volume of 100 μl for 10 min. The reaction was terminated and cells fixed by aspirating each well and adding 100 μl of 4% formaldehyde PBS solution for 20 min. Cells were washed three times with 100 μl wash buffer (0.1% Triton X-100/PBS solution), followed by a 20 min incubation with 0.6% H2O2 Triton/PBS solution to quench endogenous peroxidases. After washing the cells three times again with wash buffer, and after a 1 h incubation with 10% BSA in Triton/PBS solution (to block nonspecific antibody binding), cells were incubated overnight (about 12 h) with a 1:250 dilution of PhosphoPlus® p44/42 1° antibody in the Triton/PBS solution (100 μl) containing 5% BSA at 4°C. Cells were washed three times with wash buffer for 5 min and incubated with 100 μl HRP-conjugated goat anti-rabbit 2° antibody (1:100 dilution) with 5% BSA at room temperature for 1 h. Again, cells were washed three times with wash buffer for 5 min, and then twice with PBS. Cells were then incubated with 50 μl of an o-phenylenediamine (OPD) solution (0.4 mg/ml OPD, 17.8 mg/ml Na2HPO4·7H2O, 7.3 mg/ml citric acid and 0.015% H2O2) for 15 min at room temperature in the dark. The reaction was terminated by the addition of 25 μl of 1 M H2SO4, and the well-solution analyzed spectrophotometrically (using the Vmax Kinetic Microplate Reader from Molecular Devices) at absorbance wavelengths A490A650.

AA Release Assay

Measurements of phospholipase A2 (PLA2) AA release were made by modifying a protocol described by Berg et al (1996). CHO-K1 cells (both wild type and those transfected with the human D2L receptor) were seeded in 24-well plates in Ham's F-12 media (10% FBS) at 50 000 cells/well and allowed to grow at 37°C and 5% CO2 for 24 h. A measure of 500 μl of serum-free Ham's complete media containing 0.5 μCi/ml [5,6,8,9,11,12,14,15-3H]AA was added to wells, after which cells were preincubated for 5 h. Cells were then washed three times for 5 min with Hank's balanced salt solution (HBSS) containing 0.5% fatty acid-free BSA and appropriate enzyme inhibitors and antagonists (500 μl/well/wash). Following the washes, the cells were incubated for 15 min with appropriate agonists with or without ATP dissolved in the HBSS/BSA (1 ml/well). Three 200 μl sample aliquots were taken from each well and the radioactivity of the samples was counted using liquid scintillation spectrometer techniques.

Epifluorescence Microscopy

HEK293 cells had been stably transfected with FLAG epitope-tagged D2L receptors, and their functional integrity confirmed previously (Vickery and von Zastrow, 1999). As described previously (Vargas and von Zastrow, 2004), these cells were plated on glass coverslips and the surface receptors specifically labeled using 3 μg/ml anti-FLAG M1 monoclonal antibody. Cells were exposed to the indicated ligands at 37°C for 30 min, fixed using 4% formaldehyde dissolved in PBS, and then labeled receptors were detected by secondary incubation with Cy3-conjugated goat anti-mouse antibody (1:1000 dilution). Specimens were visualized by confocal fluorescence microscopy using a Zeiss LSM510 microscope fitted with a Zeiss 63XNA1.4 objective operated in single photon mode with standard filter sets and standard (1Airy disc) pinhole. Metamorph software (Molecular Devices) was employed to count internalized vesicles in 20 random cells per condition per coverslip, with three to six coverslips representing each condition.

Data Analysis

Except where noted, data are expressed as means±SEM. Data from receptor saturation isotherms was analyzed using nonlinear regression with a one-site hyperbolic model, and the data converted to KD and Bmax. The receptor competition data were analyzed by nonlinear regression using a sigmoidal model with variable slope, yielding IC50 (concentration inhibiting 50% of total binding IC50) and Hill slopes values. The IC50's were corrected for radioligand concentration and converted to concentration corrected IC50 (apparent affinity constant) when Hill coefficient (nH)≠1.0 (K0.5) values using the Cheng–Prusoff formula for a bimolecular competition model (Cheng and Prusoff, 1973). Changes in affinity in the competition assays were tested for significance by performing a one-way analysis of variance (ANOVA), followed by a Tukey–Kramer multiple comparisons post hoc analysis. Functional dose–response curves also were analyzed by nonlinear regression using Prism 4.0's sigmoidal equation with variable slope in order to determine estimates of intrinsic activity and apparent potency. Differences in potency values between the MAPK and AA release effectors for each agonist was analyzed with an unpaired two-sided t-test, whereas differences among the drug groups were assessed by ANOVA. Significant ANOVA results were followed by the appropriate (Bonferonni's or Dunnett's) post hoc analysis, again performed using Prism 4.0.


[3H]N-Methylspiperone Binding

The saturation data for [3H]N-methylspiperone binding to this receptor fit a one-site binding model in both cell lines, and yielded the following parameters. (1) CHO-hD2L: KD=0.39±0.11 nM, Bmax=6.57±0.63 pmol/mg protein; and (2) HEK293-FLAG-D2L: KD=0.34±0.07 nM, Bmax=1.5±0.26 pmol/mg protein (N=3 for both lines).

Competition of four D2 ligands vs [3H]N-methylspiperone binding was carried out using CHO-hD2L cell membranes both in the absence and presence of 600 μM GTP. The resulting competition curves were best fitted to a variable slope binding model, from which their apparent affinities (K0.5) and Hill slope values (nH) were derived (Table 1). ANOVA was utilized to evaluate shifts in affinity in the presence of GTP. Both quinpirole and (−)3PPP demonstrated significant loss of affinity in the presence of GTP, as would be expected of a GPCR agonist, and is illustrated by a rightward shift of their competition curves (Figure 1a and b). They also display Hill slope values typical of agonists (nH<1) in the absence of GTP. Conversely, aripiprazole failed to demonstrate a significant change in affinity (Table 1 and Figure 1c). It was observed, however, that there was a small, but consistent, shift in every experiment. A more stringent analysis (paired one-tailed t-test) was employed, and it determined that the slight shifts were indeed significant. It appears, therefore, that the binding of aripiprazole is slightly affected by the G-protein-coupled state of the D2L receptor, and is unique among the D2 receptor agonists studied. Moreover, the Hill slope for aripiprazole (nH>1) does not correspond with what would be expected of either an agonist or antagonist following the law of mass action. There appears to be a suggestion of positive cooperativity that affects the binding of aripiprazole to the D2L receptor. As expected, the apparent affinity of the typical APD and known D2 receptor antagonist, haloperidol, was not affected by GTP (Figure 1d), and its Hill slope value in the absence of GTP corresponds well with what would be expected of an antagonist (nH1).

Table 1 Binding Affinities and Hill Slope Values of Ligands for hD2L Receptors in CHO Cells in the Absence and Presence of 600 μM GTP
Figure 1
figure 1

Binding of test ligands to hD2L receptors in the presence or absence of 600 μM GTP. Membranes were prepared from CHO cells stably expressing hD2L receptors. Receptors were labeled with 0.3 nM [3H]N-methylspiperone. Nonspecific binding was defined by 10 μM haloperidol. For each compound, five to seven independent experiments were performed. The curves shown are from representative experiments. Summary of quantitative analysis is shown in Table 1.

Aripiprazole Activation of D2 Receptor-Mediated MAPK Phosphorylation

To confirm that the D2 receptor-mediated activation of the MAPK pathway was indeed regulated by MEK, the maximal MAPK phosphorylation by quinpirole was shown to be fully inhibited by 50 μM PD98059, an inhibitor of MEK (data not shown). Quinpirole was found to exhibit slightly higher intrinsic activity than the endogenous agonist dopamine, while (−)3PPP and aripiprazole displayed about 60 and 50% of the activity of quinpirole, respectively (Figure 2). The rank order of potency among the four compounds was dopamine=quinpirole>(−)3PPP>aripiprazole. Haloperidol was able to fully inhibit maximally stimulating concentrations of all four agonists, further indicating that the pathway is D2 receptor mediated. It was also confirmed that aripiprazole failed to promote the phosphorylation of MAPK in the untransfected CHO-K1 cell line (Figure 3). The other atypical APDs (clozapine, olanzapine, amisulpride, and melperone) showed no significant intrinsic activity for MAPK phosphorylation, and largely inhibited a maximally stimulating concentration (100 nM) of quinpirole (Figure 3). The inability of clozapine, olanzapine, and melperone to inhibit the MAPK effector pathway fully can be attributed to inadequate fractional occupancy at concentrations used rather than low intrinsic partial agonism.

Figure 2
figure 2

Activation of MAPK phosphorylation. The ability of D2L ligands with known agonist activity to phosphorylate MAPK was assessed in CHO-hD2L cells, and observations were quantified using an enzyme-linked immunosorbent assay. Cells were incubated with increasing concentrations of D2L ligand for 10 min at room temperature (20°C). Both quinpirole and dopamine displayed intrinsic activities that corresponded to full agonists, while aripiprazole and (−)3PPP were partial agonists. The rank order of potency was quinpirole=dopamine>(−)3PPP>aripiprazole. Data are expressed as a percentage of the maximal stimulation of quinpirole over basal phosphorylation. All values represent the mean±SEM of four to five experiments conducted in quadruplicate.

Figure 3
figure 3

Atypical APD effect on D2L-mediated phosphorylation of MAPK. To evaluate the intrinsic activity of haloperidol (Hal), amisulpiride (Ami), clozapine (Clz), melperone (Mel), olanzapine (Ola), and aripiprazole (Ari), the following design was used. The open bars show the intrinsic activity of each compound alone, where the compounds were used at a concentration of 10 μM. The antagonism study (black bars) also used this same concentration of each potential antagonist vs a challenge concentration of quinpirole (100 nM except in the case of aripiprazole where 10 μM was used (see Results)). None of the compounds except aripiprazole caused a significant response alone. All of the atypical APDs except aripiprazole were able to block quinpirole stimulation to a similar degree as haloperidol. Inset: The degree of MAPK stimulation elicited by aripiprazole (Ari) (closed bars) relative to basal levels of MAPK activity (open bars) in the untransfected parental CHO-K1 cell line. Data are expressed as a percentage of the maximal stimulation of quinpirole over basal phosphorylation. Each value represents the mean±SEM of three independent experiments conducted in triplicate.

Aripiprazole Promotes the D2 Receptor-Mediated Potentiation of AA Release

There has been much debate on the pathway mechanism of GPCR-mediated AA release, and it appears that the mechanism may vary somewhat among different receptors and may be dependent on multiple signaling pathways (Xu et al, 2002). We used staurosporine (a pan-kinase inhibitor), PD98059 (an MEK inhibitor), and Ro318220 (a protein kinase C (PKC) inhibitor), in the presence of 10 μM quinpirole. Staurosporine and PD98059 both inhibited AA release (54 and 73%, respectively), but did not cause an additive change when cells were treated with both (53% inhibition). Analysis by ANOVA with Bonferroni's post hoc test demonstrated that there was no significant difference among these three inhibitor treatment scenarios (Figure 4). Treatment with Ro318220 failed to affect the intrinsic activity of quinpirole, indicating that the activity of PKC does not appear to influence D2L-meditated potentiation of AA release in this CHO cell line. Our observations indicate that, at least in the CHO cell line, the potentiation of AA release as regulated by hD2L receptor activation is dependent on multiple pathways. Mepacrine (100 μM) fully inhibited the quinpirole potentiation of AA release, whereas melittin was found to stimulate the release of AA (data not shown), suggesting the role of PLA2 in the hD2L receptor-mediated potentiation of the AA release signaling pathway.

Figure 4
figure 4

PD98059 and staurosporine inhibition of [3H]AA release. Cells were treated with PD98059 (50 μM) and staurosporine (100 nM) either alone or together in the presence of a maximally stimulating concentration of quinpirole (10 μM). Both inhibitors were able to block significantly quinpirole-stimulated [3H]AA release, but there was no significant difference used alone or together (ANOVA, Bonferroni post hoc). In addition, neither alone nor in combination did these compounds cause total inhibition of [3H]AA release. Ro318220 (100 nM) was also used in the presence of quinpirole, but this PKC-specific inhibitor failed to block quinpirole activity. Data are expressed as a percentage of the maximal stimulation of quinpirole over ATP basal [3H]AA release. Each value represents the mean±SEM of three independent experiments conducted in triplicate.

Similar to the MAPK effector studies, all four of the D2 ligands were able to potentiate the release of AA to varying degrees. Both quinpirole and dopamine fully stimulated AA release, while aripiprazole and (−)3PPP demonstrated partial activity of the effector pathway (Figure 5). The rank order of potency among the four compounds was aripiprazole>dopamine=quinpirole>(−)3PPP. Of interest, however, were the potencies that characterized the activities of the compounds for the pathway (Table 2). While the three typical agonists had potencies for AA release fairly congruent with those observed for the MAPK pathway, aripiprazole illustrated a 20-fold higher potency for the D2-mediated potentiation of AA release than MAPK phosphorylation. Further analysis of the potency differences between MAPK and AA release experiments for each drug using ANOVA (followed by a Dunnett's post hoc multiple comparisons test) illustrated that only aripiprazole exhibited a significant potency shift relative to that of quinpirole (p<0.01). Incidentally, the EC50 of aripiprazole for AA release is similar to the potency of aripiprazole reported for D2 receptor-mediated inhibition of AC (Lawler et al, 1999; Burris et al, 2002; Shapiro et al, 2003). As with the MAPK studies, haloperidol was able to fully inhibit the ability of all four compounds to potentiate AA release, verifying D2L receptor activation as the means by which these ligands were able to exact their influence on the pathway. It was also observed that aripiprazole was unable to promote the release of AA in the parental CHO-K1 cell line, confirming that the activity of aripiprazole at this effector is specific to the transfected D2L receptor. Finally, none of the other atypical APDs examined illustrated agonist activity for the AA release pathway (Figure 6).

Figure 5
figure 5

Potentiation of [3H]AA release. The ability of D2L ligands with known agonist activity to potentiate the release of [3H]AA was assessed in CHO-hD2L cells that were incubated with 0.5 μCi/ml [3H]AA-supplemented media for 5 h. Cells were then exposed to varying concentrations of ligand at 37°C for 15 min dopamine stimulated AA release with similar intrinsic activity to quinpirole, while both (−)3PPP and aripiprazole were partial agonists. The rank order of potency was aripiprazole>quinpirole=dopamine(−)33PPP. Data are expressed as a percentage of the maximal stimulation of quinpirole over ATP basal [3H]AA release. All values represent the mean±SEM of three independent experiments conducted in triplicate.

Table 2 Functional Potencies of Agonists for hD2L Receptors in CHO Cells
Figure 6
figure 6

Atypical APD effect on D2L receptor potentiation of [3H]AA release. To evaluate the intrinsic activity of haloperidol (Hal), amisulpiride (Ami), clozapine (Clz), melperone (Mel), olanzapine (Ola), and aripiprazole (Ari), the following design was used. The open bars show the intrinsic activity of each compound alone, where the compounds were used at a concentration of 10 μM. The antagonism study (black bars) also used this same concentration of each potential antagonist vs a challenge concentration of quinpirole (100 nM except in the case of aripiprazole where 10 μM was used (see Results)). None of the compounds except aripiprazole caused a significant response alone. All of the atypical APDs except aripiprazole were able to block quinpirole stimulation to a similar degree as haloperidol. Inset: The degree of MAPK stimulation elicited by aripiprazole (closed bars) relative to basal levels of MAPK activity (open bars) in the untransfected parental CHO-K1 cell line. Data are expressed as a percentage of the maximal stimulation of quinpirole over ATP basal [3H]AA release. Each value represents the mean±SEM of two to three independent experiments conducted in triplicate.

Aripiprazole Fails to Stimulate Internalization of the FLAG-Tagged hD2L Receptor

It was originally thought a compound that displayed an agonist functional profile would induce some degree of receptor endocytosis, and that competitive antagonists, by their very nature, would be unable to provoke such a response (Ferguson, 2001). More recent evidence indicates, however, that certain antagonists can internalize certain receptors (Berry et al, 1996; Roettger et al, 1997), and that the internalization process can be agonist-specific depending on the cellular milieu specific to that receptor expression system (Ryman-Rasmussen et al, 2005). These observations complement the fundamentals of functional selectivity, which state that the intrinsic activity of a ligand is dependent on the pathway being examined and the model around which the investigation revolves. Thus, the study of the ability of a ligand to stimulate receptor internalization has been utilized to understand this aspect of the functional profile of many GPCR ligands. Utilizing HEK293-fD2L cells (Bmax=1.50±0.26 pmol receptor/mg protein) and epifluorescent microscopy techniques, we were able to observe a pronounced dopamine-induced increase in the internalization of the FLAG-tagged D2L receptors relative to basal levels (Figure 7a and b). Neither aripiprazole, nor the D2 partial agonist (−)3PPP, however, were able to elicit a detectable endocytotic response relative to basal levels (Figure 7c and d). Quantification of these results is shown in Figure 8.

Figure 7
figure 7

Ligand effects on receptor internalization. FLAG-tagged D2L receptors expressed stably in HEK293 cells were specifically labeled at 37°C with M1mAb as described in Materials and methods. Fluorescence microscopy was used to visualize the localization of antibody-labeled receptor in cells incubated for 30 min in: (a) No drugs; (b) 10 μM dopamine; (c), 10 μM aripiprazole; or (d) 10 μM (−)3PPP. Representative micrographs of each condition are shown.

Figure 8
figure 8

Quantification of ligand-mediated differences in dopamine receptor internalization in stably transfected D2L cells. Ligand-dependent dopamine receptor internalization was assayed as described in the Materials and methods. A significant number of endocytic vesicles containing endocytosed M1 antibody were observed in untreated cells (NT), consistent with the constitutive internalization that has been observed for this receptor. There was, however, a large increase in antibody uptake induced by dopamine (10 μM for 30 min). In contrast, neither aripiprazole (Arip) or (−)3PPP-mediated significant endocytosis of monoclonal antibody. The bars represent the mean number of antibody-positive vesicles (±SE) detected by MetaMorph (Molecular Devices) analysis of a region of interest that outlined the entire intracellular area in 20 cells.


Pharmacotherapy for the treatment of schizophrenia has utilized several strategies over the past fives decades, although the pharmacological mechanism common to all successful drugs is the antagonism of dopamine D2 receptors (Kapur and Mamo, 2003). The unique pharmacology of aripiprazole, a drug having both partial agonist and antagonist activity at D2 receptor functions depending on the end point under study, suggests that functionally selective ligands may provide a new arena for the development of novel therapeutics for psychoses and other disorders (Miyamoto et al, 2000; Davies et al, 2004; Maudsley et al, 2005). The current study was designed to shed further light on the question of whether the D2 activity of aripiprazole is simple partial agonism or a case of ligand-induced differential signaling. To further investigate this issue, the D2 receptor-mediated signaling profile of aripiprazole was compared to several other atypical APDs and to both a full and partial agonist, by investigating effectors that, to date, had been used infrequently to characterize APD pharmacology.

First, the apparent affinity of the D2 receptor ligands was determined in CHO cells stably expressing the hD2L receptor. The rank order of apparent affinities was similar to what has been reported previously (haloperidol>aripiprazole>quinpirole>(−)3PPP). Agonists for GPCR receptors are expected to produce competition isotherms that shift to the right in the presence of guanine nucleotides (Lefkowitz et al, 1978), a phenomenon clearly illustrated by the current quinpirole and (−)3PPP binding data. Although GTP caused a consistent shift in the aripiprazole competition curves, this effect was very small, and was statistically significant only when utilizing a one-sided paired t-test. Conversely, quinpirole and (−)3PPP, but not haloperidol, caused a larger shift in affinity and Hill slope (see Table 1). These data with aripiprazole may reconcile with the previously published GTPγS binding data in which aripiprazole was found to be a pure antagonist of D2 receptor-mediated GTPγS release in CHO-hD2L membranes (Shapiro et al, 2003).

Also of interest were Hill slopes (nH) for aripiprazole that were greater than 1.0. A Hill slope less than one is often considered a reflection of the ability of a ligand to bind to G protein-precoupled receptors with a higher affinity than uncoupled receptors. Thus, antagonists tend to have Hill slope values close to one since they fail to differentiate between the precoupled and uncoupled receptor populations, thereby obeying the general law of mass action for single site competition. As expected, both the typical full and partial agonists quinpirole and (−)3PPP had different affinities for these populations of hD2L receptors based on the changes in competition curves seen in response to GTP. The competition isotherm of haloperidol fits the binding profile of a typical antagonist (ie it both lacks a GTP shift and has an nH value close to one). The steep slope for aripiprazole, as well as a small but significant GTP effect, suggests that additional mechanisms are involved. One hypothesis for such observations is that certain ligands may promote positive cooperativity of dimerized or oligomerized receptor (Lavoie and Hebert, 2003), or by allosteric interaction with a secondary domain of the receptor or another protein present in the microdomain (eg a scaffolding or accessory/chaperone protein). It is plausible that the receptor–receptor or receptor–protein interaction that imparts this positive cooperativity only occurs upon the induction of specific receptor conformations. Whatever the case, the unique receptor binding characteristic of aripiprazole is not exclusive to the CHO-hD2L stable cell model, as similar observation was made in brain tissue (Lawler et al, 1999). Further study is necessary to understand the underlying mechanism.

It is known that D2 receptor agonists can stimulate the MAPK pathway in stable C6 and CHO-transfected cells (Luo et al, 1998; Choi et al, 1999). By modifying an ELISA high-throughput assay (Versteeg et al, 2000), we were able to determine intrinsic activities and potencies for the D2 ligands we studied. The intrinsic activity of aripiprazole correlates well with its reported activity for the D2L-mediated inhibition of cAMP accumulation (Lawler et al, 1999; Burris et al, 2002; Shapiro et al, 2003). The major point of interest, however, lies in the low potency of aripiprazole for the MAPK effector pathway. With the exception of aripiprazole, prior cAMP inhibition EC50 data for all compounds correspond fairly well with the MAPK potencies reported in this study (Wilson et al, 2001; Burris et al, 2002; Gay et al, 2004).

The biosynthesis and regulation of prostaglandin-like compounds by catecholamines has been of interest for several years (Levine and Moskowitz, 1979). The ability of the D2 receptor to mediate the release and metabolism of AA from the intercellular membrane is known (Piomelli et al, 1991; Felder et al, 1991), and although the exact mechanism is not well understood, there has been much discussion as to how it might be regulated by other signaling effectors, most notably AC, MAPK, and PKC (see Chakraborti, 2003 for a review). There appears to be significant evidence that the GPCR-mediated release of AA, in fact, may be regulated by more than just one pathway, and both the MAPK and PKC effectors appear to play significant roles in AA release (Xu et al, 2002). For this study, it was vital to demonstrate that the release of AA was not solely dependent on a D2 receptor-mediated effector already under study (ie MAPK phosphorylation). Neither staurosporine nor PD98059 was able to inhibit the D2 receptor-mediated release of AA fully. This demonstrates a significant degree of independence of AA release from modulation by MAPK signaling pathways. In addition, PKC inhibition had no effect on the D2 receptor regulation of this pathway. This independence of D2L-mediated AA release from PKC indicates that this cell line differs from some others (Xu et al, 2002).

The intrinsic activity (Emax) of aripiprazole for AA release proved to be similar to its ability to stimulate the phosphorylation of MAPK (Figures 3 and 6), although its potency for the former end point clearly paralleled its reported potency for the D2 receptor-mediated inhibition of cAMP accumulation (Lawler et al, 1999; Burris et al, 2002; Shapiro et al, 2003). This method of comparison assessment is valid because the intrinsic activity of aripiprazole was similar at both MAPK and AA release, and therefore the direct comparison of ED50's is equivalent to the direct comparison of ED50/Emax. Unlike aripiprazole, the D2 agonists quinpirole, dopamine, and (−)3PPP all displayed potencies for the release of AA similar to that for the phosphorylation of MAPK. In fact, the difference in the potencies of aripiprazole between the two effectors is over 20-fold (Figure 9). This disparity is especially striking in context with the observation hat other functionally selective D2L ligands (eg DHX, DNS, RNPA) also have demonstrated significant potency differences among D2 receptor effectors (Gay et al, 2004).

Figure 9
figure 9

Relative potency of D2L ligands for the potentiation of AA release and phosphorylation of MAPK. Ligands with similar potency for both AA release and MAPK phosphorylation have a log (fold change) of close to zero (quinpirole (QUIN) and (−)3PPP), whereas compounds having decreased potency for MAPK phosphorylation relative to AA release display a negative change (aripiprazole (ARI)), and ligands with an increased potency for MAPK phosphorylation compared to AA release display a slight positive change (dopamine (DA)). Data illustrated for each ligand are expressed as the log of EC50 (nM) for MAPK phosphorylation divided by their respective log of EC50 value for AA release.

The intrinsic activity of aripiprazole for the D2L receptor-mediated phosphorylation of MAPK may not be of significant physiological relevance since the cell system used had a high level of receptor expression (Bmax6 pmol/mg protein). It is well documented that, in systems with high receptor reserve, compounds with low intrinsic activity can mimic compounds with higher intrinsic activities for that effector system (eg Watts et al, 1995). This notion, coupled with preliminary experiments that utilized the nonspecific, irreversible protein binding compound EEDQ to reduce receptor levels (data not shown), suggests that D2 modulation of MAPK may not be an effector pathway affected by aripiprazole in physiologically relevant systems. It underscores the importance of using, when available, cell lines that mirror as much as feasible those cells expressing the receptor under study in situ.

The model commonly used to illustrate GPCR internalization describes an agonist–receptor interaction that promotes receptor phosphorylation via a GRK or other kinase, with subsequent arrestin binding and receptor sequestration via clathrin-coated pits (von Zastrow, 2003). Not all GPCRs follow the same mechanism of internalization, and the intracellular loops of a receptor, as well as the local cellular milieu, play an important role in the mechanism of endocytosis (Ferguson, 2001). D2 receptors have been shown to internalize in a manner dependent upon GRKs, clathrin, and dynamin (Kim et al, 2001), although dynamin-independent internalization can be observed in cells expressing GRKs at lower levels (Vickery and von Zastrow, 1999). Studies in an HEK293-fD2L system showed that the D2L receptor displays both basal (constitutive) and dopamine-induced internalization, and that haloperidol can block dopamine-induced, but not constitutive, internalization (Vickery and von Zastrow, 1999). Our study indicates that neither aripiprazole nor the typical partial agonist (−)3PPP induce a significant degree of internalization relative to basal levels, indicating that either D2 ligands with low intrinsic activity fail to produce D2 receptor internalization or that these two compounds are unique in their promotion of receptor conformations that cannot be phosphorylated by the appropriate kinases. Although (−)3PPP can desensitize the D2 receptor (Lahti et al, 1998), it is clear that this does not necessarily predict internalization (Lewis et al, 1998; Ryman-Rasmussen et al, 2005). Further experiments are needed to determine whether D2 receptor internalization is a product unique to ligands that are full agonists, or if aripiprazole and (−)3PPP are unique in their inability to internalize the D2 receptor.

Some of the most efficacious atypical APDs (clozapine, amisulpride, olanzapine), as well as a compound considered to have good atypical APD potential (melperone), were also studied. Vanhauwe et al (2000) had reported that these compounds should be considered as D2 receptor antagonists based on their inhibition of AC, although melperone was not studied by them. We hypothesized that a study of their regulation of D2L function might reveal functional characteristics that explained some of their behavioral atypicality. The current study demonstrates that none of these four atypicals have any D2L intrinsic activity for either MAPK phosphorylation or AA release, and all four blocked the activity of quinpirole for both endpoints. Thus, these data suggest that neither the efficacy nor low EPS of these compounds involves functional selectivity.

In conclusion, the promise of even more effective D2 partial agonists has been widely proposed based on the success of aripiprazole (Burris et al, 2002; Lieberman, 2004; Bolonna and Kerwin, 2005). Conversely, we have suggested that functional selectivity at D2 receptors, probably combined with actions at nondopamine receptors (Lawler et al, 1999; Shapiro et al, 2003), is the more likely mechanism responsible for the atypicality of this drug. It seems clear to us that the available evidence (including the current work) supports only the latter hypothesis. No other D2 partial agonist has shown similar therapeutic promise to aripiprazole. More importantly, even those who have advocated for simple partial agonism (Burris et al, 2002) now have shown cell-dependent differences in the intrinsic activity of aripiprazole (Tadori et al, 2005). The most parsimonious way to reconcile the available data is accept the hypothesis that the pattern of D2 functional selectivity, and/or combined with actions at other receptors systems (eg 5-HT1A, 5-HT2C, etc), mediate the novel actions of aripiprazole, rather than simple partial agonism. More detailed understanding of these mechanisms may help find a drug with an improved clinical profile.